[unreadable] Prostate cancer is the most common malignancy of men in the U.S. and there is an urgent need to develop signatures that distinguish clinically significant disease with life threatening potential from the more common "indolent disease". We have developed methods for determining cell-specific gene expression of prostate carcinoma for four major cell types or groups within prostate carcinoma, including tumor cells, BPH, stroma, and cystic atrophy, and have identified over 3900 genes that are uniformly expressed in one cell type or another. The approach has been extended to identify ~1100 genes specifically associated with relapse as a surrogate for aggressive prostate cancer. These genes are lists by bioinformatics (Aim 1) and experimental methods (Aims 2-4) to develop a predictive signature and to validate the signature in a prospective trial. A consortium of strategic partners will cooperate in completing this goal. We will work with Jeffrey Trent of TGen to analyze 100 fresh surgical samples by quantitative PCR for 200 prioritized genes (Aim 2). Strategic partners at the University of California at San Diego and at Irvine will supplement the SKCC/Sharp HealthCare collection of clinically annotated paraffin blocks of prostate cancer to complete a current project to make a 1000 case tissue microarray (TMA). This TMA will be examined with 200 prioritized antibodies, manufactured by a commercial strategic partner, in cooperation with John Reed of the Burnham Institute (Aim 3). These arrays are specifically designed to (i) correlate gene expression with survival in a retrospective study, (ii) test whether stroma specific genes are correlated with adverse clinical outcomes, and (iii) whether tissue removed prior to the development of prostate cancer expresses predictive genes. Genes that are successfully validated by these methods will be utilized in a prospective observational clinical trial to be carried out at three sites (the Northwestern University Medical School Prostate SPORE, UCSD, and SKCC/Sharp HealthCare) to test the ability of the signatures to accurately predict early relapse (Aim 4). Changes in genes that are predictive of clinically relevant events will have strong justification for further development as predictive biomarkers. [unreadable] [unreadable]